2008AJ....136..840J

Using new spectra of 88 K- and M-type subdwarfs, we consider novel methods for assigning their spectral types and take steps toward developing a comprehensive spectral sequence for subdwarf types K3.0 to M6.0. The types are assigned based on the overall morphology of spectra covering 6000-9000 Å. The types and sequence presented link the spectral types of cool subdwarfs to their main-sequence counterparts, with emphasis on the relatively opacity-free region from 8200-9000 Å. When available, supporting abundance, kinematic, and trigonometric parallax information is used to provide more complete portraits of the observed subdwarfs. We find that the CaHn (n = 1-3) and TiO5 indices often used for subdwarf spectral typing are affected in complicated ways by combinations of the subdwarfs' temperatures, metallicities, and gravities, and we use model grids to evaluate the trends in all three parameters. Because of the complex interplay of these three characteristics, it is not possible to identify a star as an "extreme" subdwarf simply based on very low metallicity, and we suggest that the modifiers "extreme" or "ultra" only outline locations on spectroscopic indices plots, and should not be used to imply low or very low metallicity stars. In addition, we propose that "VI" be used to identify a star as a subdwarf, rather than the confusing "sd" prefix, which is also used for hot O and B subdwarfs that are unrelated to the cool subdwarfs discussed in this paper.